Pith Number

pith:GY43OMZ3

pith:2026:GY43OMZ3V66BX2ECGX6W2HOQP4

not attested not anchored not stored refs resolved

Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning

Changwen Zheng, Huijie Guo, Hui Xiong, Jiahuan Zhou, Jingyao Wang, Peizheng Guo, Wenwen Qiang

Treating multiple reasoning paths for one question as counterfactual experiments trains LLMs to favor stable and transferable reasoning patterns over lucky guesses.

arxiv:2602.06475 v2 · 2026-02-06 · cs.LG

Open paper page JSON Open Graph Bundle Merged state What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{GY43OMZ3V66BX2ECGX6W2HOQP4}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

We propose Group Causal Counterfactual Policy Optimization to explicitly train LLMs to learn generalizable reasoning patterns. It proposes an episodic causal counterfactual reward that jointly captures (i) robustness, encouraging the answer distribution induced by a reasoning step to remain stable under counterfactual perturbations; and (ii) effectiveness, enforcing sufficient variability so that the learned reasoning strategy can transfer across questions.

C2weakest assumption

That multi-candidate reasoning trajectories for a fixed question can be validly interpreted as a family of counterfactual experiments with sufficient theoretical support, and that the resulting robustness and effectiveness reward will produce reasoning patterns that generalize without introducing new failure modes or biases.

C3one line summary

Group Causal Counterfactual Policy Optimization trains LLMs on generalizable reasoning by defining episodic rewards for counterfactual robustness and transferability then optimizing the policy with token-level advantages.

References

19 extracted · 19 resolved · 8 Pith anchors

[1] Training language models to reason efficiently

[2] Evaluating Large Language Models Trained on Code · arXiv:2107.03374

[3] Pass@ k training for adaptively balancing exploration and exploitation of large reasoning models

[4] Training Verifiers to Solve Math Word Problems · arXiv:2110.14168

[5] Group causal policy optimization for post-training large language models.arXiv preprint arXiv:2508.05428,

Receipt and verification

First computed	2026-05-18T03:09:23.754563Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

3639b7333bafbc1be88235fd6d1dd07f24fd3d19aaa6381bb3febc6451ec41e9

Aliases

arxiv: 2602.06475 · arxiv_version: 2602.06475v2 · doi: 10.48550/arxiv.2602.06475 · pith_short_12: GY43OMZ3V66B · pith_short_16: GY43OMZ3V66BX2EC · pith_short_8: GY43OMZ3

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/GY43OMZ3V66BX2ECGX6W2HOQP4 \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 3639b7333bafbc1be88235fd6d1dd07f24fd3d19aaa6381bb3febc6451ec41e9

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "f2b0b5545aeada88c5c61f59a3fe22d06cfb6430a1b9d68a2b9b87d21a91f7dd",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-02-06T08:03:11Z",
    "title_canon_sha256": "6ff29da67c0f21410f70c05d05895ed21596719fc205159b6d48e4dd92eb0618"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2602.06475",
    "kind": "arxiv",
    "version": 2
  }
}